Process optimization in a supervision support system

Abstract

We present details of the optimization task in a real-time (Thompson and Mertz 1993) knowledge-based (Moore et al. 1991, Larsson 1992) supervision support system in the coal washing domain. The Ash Control Model (AshMod) assists operators in maximizing clean coal yield while keeping ash (impurity) content within acceptable limits. AshMod assists the operator in plant situation assessment, fault diagnosis, and performance optimization. Situation assessment and fault diagnosis are mentioned briefly, since they have been described elsewhere in detail (Villanueva and Lamba 1997, 1998). We focus on the optimization task, which employs a hybrid artificial intelligence and operations research approach. The process is modelled through a set of extended states associated with the entire process and with individual components (circuits) within the plant. The scheduling optimizer continuously monitors the process, assesses the process state, and dynamically plans and performs integer and real optimization of a sequence of actions. The supervision support system captures domain knowledge through multiview models (Terspra et al. 1993) such as the goal tree success tree, plant schematic and fault cause network. AshMod performs deep reasoning through the use of knowledge models (Lind 1994) that capture purpose, function, structure, behaviour and heuristics. These knowledge models are mentioned briefly, since they have also been described elsewhere in detail (Villanueva and Lamba 1997, 1998). The supervision support system (Lamba 1995) is currently undergoing online validation at the B&C Coal Washing Plants operated by the Broken Hill Proprietary Limited (BHP 1998) at Port Kembla, Australia. The system is expected to be fully operational by the end of 1998.